![](images/null.gif) |
![](images/null.gif) |
![](images/null.gif) |
|
Prevalence of Three-class Resistance in North Carolina cohort study; K65R & tenofovir response, and predicting salvage response to fosamprenavir/r
|
|
|
Mark Mascolini
Report #2
XV Intl HIV Drug Resistance Workshop
June 13-17, 2006, Sitges, Spain
Prevalence of resistance among antiretroviral-experienced people- and which mutations favor a good response to the next regimen- got plenty of attention at the Resistance Workshop. A cohort study at the University of North Carolina and a mountainous database analysis both saw hopeful signs on resistance trends in the United States. Two studies- and the database inquest- yielded provocative findings about mutations conferring resistance to tenofovir.
Predictors of triple-class resistance
Resistance to the first three antiretroviral classes- nucleoside and nonnucleoside reverse transcriptase inhibitor (NRTIs and NNRTIs) and protease inhibitors (PIs) - proved relatively rare in a retrospective study of 1587 adults at the University of North Carolina, Chapel Hill [1]. Joseph Eron reported that no one in this cohort who started therapy with a ritonavir-boosted PI had triple-class resistance at the time of his analysis.
Defining triple-class resistance as one or more mutations conferring resistance to one or more drugs in each of the first three antiretroviral classes, Eron figured an overall prevalence of 8% in 1587 people, half of whom began treatment with a potent regimen (HAART), 20% of whom began with substandard combination therapy, and 30% of whom started with one-drug therapy. Among 752 people who tried at least one drug from all three classes, 16% had three-class resistance.
Resistance to three classes reached higher rates in people who began treatment with only one or two antiretrovirals, an expected findings since the vast majority of people who started therapy this way picked up mutations with every failure- and few escaped failure of these wimpy regimens. Among 798 people who first took one or two antiretrovirals, triple-class resistance struck 12%. In 447 people who began therapy in this fashion and worked their way through all three classes, the triple-class resistance rate reached 21%.
In contrast, only 3% of 789 people lucky enough to start therapy with HAART got triple-class resistance, including only 8% of those who started with HAART and tried all three classes. Most people in this group, 87.5%, began therapy with an unboosted PI. No one who pinned their first HAART on a ritonavir-boosted PI ended up with virus resistant to all three classes.
A long list of variables made triple-class resistance more likely, including lower pretreatment CD4 count, higher peak viral load, a pretreatment AIDS diagnosis, earlier year of first treatment, starting with a non-HAART regimen, a higher number of antiretrovirals tried, NNRTI experience, PI experience, and three-class experience.
But in a multivariate analysis, only three factors stood out as independent predictors of tripartite resistance (Table 1). For all patients studied, experience with more antiretrovirals and a non-HAART regimen predicted three-class resistance; and for people who took HAART when they started therapy, more antiretrovirals on the docket and a higher peak viral load predicted triple resistance.
![Predictors-1.gif](../images/062306/Predictors-1.gif)
Why prior ddI or abacavir threatens response to tenofovir
Using an allele-specific PCR assay that ferrets out resistant virus missed by standard assays, Michael Miller and Gilead colleagues made two important findings [2]:
- Treatment with abacavir or ddI can breed low levels of K65R, the main mutation linked to resistance to tenofovir.
- The L74V/I mutation, also evoked by ddI or abacavir, also threatens response to tenofovir.
Miller used the PCR assay to trawl for K65R in viral isolates from treatment-experienced people enrolled in a study that randomized them to add only tenofovir or placebo to an incompletely suppressive regimen [see note 3]. The viruses analyzed came from 78 people who had taken ddI or abacavir and 30 who hadn't. In the ddI-or-abacavir-treated group, 44 had the L74V mutation, 5 had K65R by standard genotyping, and 29 had neither K65R nor L74V. The group still naive to ddI and abacavir did not have L74V or K65R.
Allele-specific PCR confirmed K65R in all 5 ddI-or-abacavir-treated people in whom standard sequencing spotted that mutation. The PCR assay also disclosed low-level K65R (1.2% to 25% of the viral population) in 6 of 44 ddI-or-abacavir-treated people (14%) carrying L74V/I and in 1 of 29 ddI-or-abacavir-treated people without L74V/I. The supersensitive mutation stalker flushed out no K65R in people who had not tried ddI or abacavir.
The average 24-week viral load drop in 380 trial participants who added tenofovir measured 0.62 log copies/mL. Eight people with L74V/I but with no AZT- or d4T-related mutations (thymidine analog mutations, or TAMs) averaged a 0.31-log drop in viral load after 24 weeks of tenofovir. The entire 41-person group with L74V/I, including those with TAMs, had only a 0.17-log viral load dip after 24 weeks. Virologic responses were negligible in 8 people with K65R detected by standard genotyping and in 7 with K65R rooted out by allele-specific PCR (but see the following section for a different perspective on pre-tenofovir K65R).
A statistical analysis weighing several variables that may influence response to tenofovir linked treatment with tenofovir, higher baseline viral load, higher baseline CD4 count, and the 3TC-evoked M184V mutation with viral load drops when people added tenofovir to their regimen (Table 2). Multiple TAMs, K65R, and L74V all predicted a poor salvage response to tenofovir.
![table2-2.gif](../images/062306/table2-2.gif)
All told, low-level K65R turned up in virus from 7 of 73 patients (10%) who took ddI or abacavir before adding tenofovir to a faltering regimen in this trial. But this tenofovir-threatening mutation never showed up, even at low frequency, in nucleoside-experienced people still naive to ddI and abacavir. These occult traces of K65R predicted a poor response in the tenofovir trial, as did the ddI- and abacavir-evoked L74V mutation.
. . . but sometimes tenofovir improves response with K65R
A five-cohort US-French study presented by Andrew Zolopa (Stanford University) underscored the prickly challenge of planning rescue regimens for people with the K65R mutation [4]. In this analysis, having tenofovir in the rescue regimen improved virologic responses of people already marked by the K65R mutation, but using a thymidine analog (AZT or d4T) did not.
The study involved 129 people with K65R who tried new regimens based on genotypic resistance testing. Ninety (70%) had already used tenofovir, 93 (72%) ddI, and 54 (42%) abacavir. Seventy-three people (57%) had the M184V mutation, 37 (29%) had at least one TAM, and 13 (10%) had L74V. (The study described above [2] analyzes how L74V affects response to tenofovir.) Median viral load stood at 10,233 copies/mL before people switched to their new regimens, 102 (79%) of which included a PI and 47 (36%) of which included tenofovir.
Defining response as a sustained viral load below 400 copies/mL, Zolopa and colleagues counted 71 responders (55%). Another 22 people (17%) had a partial response, defined as at least a half-log drop in viral load. Thirty-six people had no response.
In a statistical analysis considering one factor at a time- independent of any other factor- no presalvage NRTI mutation predicted a good or bad virologic response. Neither did previous use of any antiretroviral, number of antiretrovirals in the salvage regimen, or switching to a new antiretroviral class. Only a lower presalvage viral load, a higher genotypic sensitivity score [see note 5], and treatment with AZT or d4T favored reaching a viral load below 400 copies/mL.
But in a multivariate analysis, including AZT or d4T in the new regimen did not promote a better response. Including tenofovir did favor a good virologic response, even though 70% of the study group had already taken tenofovir and everyone had the tenofovir-crippling K65R mutation. Zolopa and coworkers speculated that the group's high prevalence of M184V- which is hypersusceptible to tenofovir- may have improved responses to tenofovir in this group. But they noted that M184V did not correlate with a good response even in the univariate (one-factor-at-a-time) analysis.
Still, Zolopa proposed in follow-up e-mail to this reporter, ÒI suspect that many patients failing with K65R will still have some activity with tenofovir- our study suggests about a 0.5 logÓ drop, probably because so many people carry M184V. ÒIn fact if you look at phenotypes,Ó he explained, Òmost viruses with K65R and M184V have a fold-change in susceptibility less than 1.4- often around 1.0.Ó In other words such viruses remain highly susceptible to tenofovir. As the Gilead study [2] showed, however, other mutations carried along with a K65R mutant- most notably TAMs- can upset this equation.
A lower viral load when the new regimen began and a higher genotypic sensitivity score also independently predicted success with the rescue regimen.
Encouraging US mutation trends
Figuring out how to corral K65R mutant virus (see preceding section [4]) may be a dwindling concern of HIV treaters these days, according to results of a massive resistance database analysis by Ron Kagan of Quest Diagnostics [6].
Analyzing 173,603 viral isolates submitted to two Quest centers from 1998 through 2005, Kagan found that K65R prevalence among resistant samples peaked out in 2002-2003 and dropped back to about 4% in 2005. He speculates that trend may reflect wholesale rejection of resistance-prone triple nucleoside/nucleotide combinations including tenofovir. The K70E mutation, also a byproduct of failing tenofovir regimens, rose in recent years and correlated statistically with more tenofovir use. But prevalence of this mutation remained below 1%.
Kagan had other good news to report. First, overall resistance rates are falling in viral samples submitted to Quest. Whereas 79% of viral isolates bore resistance mutations in 1999, 46% to 52% (depending on the Quest center) carried such mutations in 2005. Triple-class-resistant virus accounted for under 20% of resistant isolates in 2005, down from 32% in 1999. Samples without AZT- or d4T-inspired TAMs jumped from 25% to 46%, probably partly because d4T gets scribbled on fewer and fewer prescription pads. Prevalence of the T69D and K70R mutations, tied to ddI failure, has dropped.
Reflecting the popularity of efavirenz, rates of NNRTI-resistant virus clambered from 46% in 1998 to 62% in 2005. Rates of the efavirenz-linked K103N mutation are up, while prevalence of Y181C, more often tied to nevirapine, is down.
Predicting salvage response to fosamprenavir/ritonavir
Bernard Masquelier (University of Bordeaux) and French national trials group (ANRS) colleagues proposed a new mutation-counting method to predict response to fosamprenavir/ritonavir in people with heavy PI experience [7].
The ANRS team based this decision tool (or algorithm) on treatment responses in 67 people who took a salvage fosamprenavir/ritonavir regimen for at least 3 months and had detectable amprenavir levels in plasma at months 1 and 3. This group had tried a median a 10 antiretrovirals (interquartile range [IQR] 8 to 12), including a median of 3 PIs (IQR 2 to 4). And they had taken PIs for a median of 8 years (IQR 6 to 9 years). When they started fosamprenavir/ritonavir (with no other PIs), their median viral load stood at 4.56 log copies/mL (about 36,000 copies/mL) and their median CD4 count at 224 cells/µL.
By treatment month 3 the median viral load fell 0.62 log copies/mL and the mean load by 1.13 log. A lengthy list of well-known protease mutations correlated with a poor response- those at positions 10, 33, 46, 54, 62, 71, 73, 82, and 90. An N88D or N88S substitution favored a better response.
Masquelier and colleagues tested a model in which 0 or 1 mutation meant no resistance, 2 or 3 mutations meant possible resistance, and 4 or 5 mutations meant resistance. Multivariate analysis rated this scoring system an independent predictor of virologic response to fosamprenavir/ritonavir. In the same analysis an N88S mutation before starting fosamprenavir/ritonavir and giving ddI with those PIs fostered a better response.
On the basis of these findings and previous work, the ANRS proposed a new algorithm for predicting response to fosamprenavir/ritonavir in people with PI experience (Table 3).
![Salvage-3.gif](../images/062306/Salvage-3.gif)
Four or five mutations among L33F, I54L/M/V/A/T/S, I62V, V82A/F/C/T/G, L90M Two or three mutations from among L33F, I54L/M/V/A/T/S, I62V, V82A/F/C/T/G, L90M
This response-predicting tool worked well in the French population studied. But Masquelier cautioned that its value must be verified in other groups.
Mark Mascolini writes about HIV infection
References and Notes
1. Napravnik S, Keys J, Quinlivan EB, et al. Prevalence and predictors of triple-class antiretroviral drug resistance in routine HIV primary. Antivir Ther 2006;11:S88.
2. Svarovoskaia ES, Margot NA, Bae AS, et al. Allele-specific PCR shows low-level K65R in treatment-experienced patients with L74V in the absence of TAMs. Antivir Ther 2006;11:S80.
3. Technical experts will want to know that a natural polymorphism (AAG) at codon 65 interferes with allele-specific PCR for K65R. The AAG polymorphism is common with subtype C HIV-1 but appeared in only 4 of 108 subtype B samples in this Gilead analysis.
4. Nevins AB, Wirden M, Rhee SY, et al. Virological response to antiretroviral therapy in the setting of the K65R mutation. Antivir Ther 2006;11:S92.
5. The genotypic sensitivity score based on the Stanford HIV resistance database adds viral susceptibility rankings for each drug in a new regimen: 1.0 gets added for a fully active drug, 0.5 for a partially active drug, and 0 for an inactive drug.
6. Kagan R, Merigan T, Winters M, et al. Newer antiretroviral treatment regimens drive HIV-1 RT and PR mutational patterns in a national reference laboratory database. Antivir Ther 2006;11:S133.
7. Masquelier B, Assoumou KL, Descamps D, et al. Genotypic determinants of the virological response to fosamprenavir/ritonavir in protease inhibitor-experienced patients. Antivir Ther 2006;11:S101.
|
|
|
![](images/null.gif) |
![](images/null.gif) |
|
|